Pharmacophore Based Screening & Modification of Amiloride Analogs for targeting the NhaP-type Cation-Proton Antiporter in Vibrio cholerae

The genome of Vibrio cholerae contains three structural genes for the NhaP-type cationproton antiporter paralogues, Vc-NhaP1, 2 and 3 mediating exchange of K and or Na for protons across the membrane. Based on phenotype analysis of chromosomal Vc-NhaP1, 2 and 3 triple deletion mutants we suggested that Vc-NhaP paralogues might play a role in the Acid Tolerance Response (ATR) of V. cholerae as it passes through the gastric acid barrier of the stomach. Comparison of the biochemical properties of Vc-NhaP isoforms revealed that VcNhaP2 is the most active among all three paralogues. Therefore, Vc-NhaP2 antiporter is a plausible therapeutic target for developing novel inhibitors targeting these ion exchangers. Our structural and mutational analysis of Vc-NhaP2 identified a putative cation binding pocket formed by antiparallel extended regions of two transmembrane segments (TMSs V/XII) along with TMS VI. Molecular Dynamics (MD) simulations suggested that the flexibility of TMSV/XII is crucial for the intra-molecular conformational events in Vc-NhaP2. In this study, we developed some putative Vc-NhaP2 inhibitors from Amiloride analogs (AAs). Amiloride is a potent inhibitor of human Na/H exchanger-1 (NHE1). Based on the pharmacokinetic properties and potential binding affinity scores we chose six AAs showing high binding affinity scores to Vc-NhaP2. In silico, the six AAs interacted with the functionally important amino acid residues located in TMSs III, IV, V, VI, VIIII and IX either from the cytoplasmic side (three AAs) or the periplasmic side (three AAs) of Vc-NhaP2. Four AAs were modified to reduce their toxicity profile compared to the original AAs. Molecular docking of the modified AAs revealed promising binding. The four selected drugs interacted with functionally important amino acid residues located on the cytoplasmic side of TMS VI, the extended chain region of TMS V and TMS XII and the loop region between TMSs VIIII and IX. Molecular dynamics simulations Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 18 February 2021 doi:10.20944/preprints202102.0408.v1


Introduction
Vibrio cholerae, the causative agent of the diarrheal disease Cholera, is considered as an 'emerging and re-emerging' disease with epidemic and pandemic potentials [1][2].V. cholerae has remarkable genetic features that enables the human pathogen to survive in different harsh environmental conditions [2].In recent decades, this enteric pathogen has developed multi-drug resistance (MDR) by acquiring numerous mobile genetic elements [3].In addition, a simulation based on an artificial neural network demonstrated that climate change might drastically elevate cholera outbreaks in the near future [4].The combination of elevated outbreaks and higher prevalence of MDR V. cholerae could lead to an irrepressible situation.Unfortunately, oral vaccination for cholera only offers temporary prevention [5].Therefore, the development of new antimicrobial treatments against this deadly pathogen is paramount.
We reported that these NhaP paralogues primarily acted as K + or Na + /H + antiporters [12][13][14][15][16][17][18].By carrying out phenotype analysis of engineered chromosomal Vc-nhaP1, Vc-nhaP2 and Vc-nhaP3 deletion mutants and complementation of each isoform, we proved that the three NhaP paralogues are essential for maintaining K + homeostasis in the cytoplasm of V. cholerae in vivo [13].Expressed in trans, neither of the Vc-NhaP paralogues was able to complement the severe potassium-sensitive phenotype of the triple-deletion mutant completely.The apo-V.cholerae had much higher survival rate compared to the triple deletion mutant, Vc∆NhaP123, when challenged by HCl (pH 3.5) [13].We therefore suggested that Vc-NhaP paralogues play a critical role in the Acid Tolerance Response (ATR) of V. cholerae as it passes through the gastric acid barrier of the stomach [13,16].The Vc-NhaP type antiporters could be a potential target to develop druggable inhibitors to treat this deadly pathogen.Notably, these NhaP-type antiporters are phylogenetically diverse and abundant in the genomes of pathogenic microorganisms [19].
As a result, specific inhibitors against NhaP-type antiporters will potentially be less damaging to benign gut microflora compared to conventional antibiotics.
Vc-NhaP2 is the most active among the three Vc-NhaP1, 2 and 3 paralogues [13].Based on mutagenesis experiments and in silico analyses, we have gathered extensive information on the structural, biological and functional properties of the Vc-NhaP2 antiporter [16][17][18].The structural analysis of Vc-NhaP2, based on the mutagenesis data, combined with the in silico structure modelling and Molecular Dynamics (MD) Simulations yielded two important elements in the organization of Vc-NhaP2: (1) a putative cation binding pocket formed by antiparallel extended regions of two transmembrane segments (TMSs V/XII) crossing each other in the middle of the membrane, and (2) a cluster of amino acid residues near the putative cationbinding pocket determining the ion selectivity [16,18].
Here, we retrieved and developed AMLRD analogues by virtual screening and rational drug design.After implementing molecular docking and MD simulations, we predict the biological consequences of the interacting AMLRD analogues.To accomplish this, we screened forty-five AMLRD analogs based on their pharmacophore-based toxicity and ADME (adsorption, distribution, metabolism, excretion) properties.Six potential AMLRD analogs were selected based on their calculated binding affinity with Vc-NhaP2.In silico, the six AMLRD analogs interacted with the functionally important amino acid residues located in TMSs III, IV, V, VI, VIIII and IX either from the cytoplasmic side or the periplasmic side of Vc-NhaP2.Three AMLRD analogs were further modified to reduce their toxicity profiles compared to the parent compound.The dynamics of the Vc-NhaP2 were further analysed by carrying Molecular Dynamics (MD) simulations and the effect on flexibility and the rigidity of the protein upon drug binding was studied.Based on our analysis we have proposed three modified drugs along with one original AMLRD analog that can be potentially used as druggable inhibitors against Vc-NhaP2.

Screening, modifications and pharmacoinformatic exploration of the potential amiloride analogs
Amiloride is a renowned diuretic capable of inhibiting cationic antiporters in various mammalian cells [22,[26][27].Forty-five chemical structures similar to AMLRD along with their Canonical SMILES and PubChem CID were initially collected from the PubChem database.For investigating ADMET properties, the values from the pkCSM server were mostly prioritized since this server uses a novel approach called graph-based signatures which outperforms other available tools [28].Data obtained from ProTox-II, OSIRIS Property Explorer and admetSAR 2.0 were also compiled in Excel spread sheets to compare the overall pharmacokinetics properties of the obtained compounds (Supplementary Excel File).

Quality of the Vc-NhaP2 antiporter structure
The generated structure of Vc-NhaP2 has 13 Transmembrane Segments (TMSs) with TMSs V and XII being discontinuous; TMSs V and XII have extended chain region that cross each other in the middle of the membrane [16,18] (Supplementary Figure 1).The discontinuous region is highly flexible and crucial for the functioning of the antiporter [16,18].The ion binding amino acid residues are mainly located in the flexible region between TMSs V, XII and TMS VI [16,18].
The final refined model of the Vc-NhaP2 channel has a MolProbity score of 0.71 with a 0 clash score and 0 bad bonds (Table 4).Moreover, 98 % of residues are located in the favored region of Ramachandran plot (Fig 2).The MolProbity score is a single number representing the one number that reflects the crystallographic resolution at which those values would be expected.
The Molprobity score of 0.71 indicates that the refined model is of a quality typically observed for high-resolution crystal structures deposited in the protein data bank with a resolution of 0.71 Å. (Table 4).The same conclusion was reached independently by ProSA-web [40] analysis that yielded a Z-score of -6.71 using only the C-α atoms of the input structure (Figure 3A).The negative Z-score value indicates that the overall model quality contained less error than typically found for native proteins of similar size as examined by X-ray diffraction.The residue score plot in Figure 3B shows negative energies as a function of amino acid sequence throughout the protein indicating few poorly defined regions in the refined Vc-NhaP2 structure (Figure 3B).

Molecular interactions between Vc-NhaP2 and selected drug candidates
Interactions between Vc-NhaP2 and the six nominated drug compounds reveal two potential drug-binding sites in the antiporter.One binding site is located on the periplasmic side of the protein and the other is on the cytoplasmic side.3-amino-N-carbamimidoyl-6-chloro-5-(dimethylamino) pyrazine-2-carboxamide (PubChem CID 137630036), AmiKol 1 and 5-amino- show potential binding at the amino acid residues located at the cytoplasmic region of TMSs IV, V and X. ADV and Achilies showed average binding affinity scores of -5.9 kcal/mol, -6 kcal/mol and -5.3 kcal/mol respectively for these three drugs (Figure 4).Noticeably, except AmiKol 1, all the drugs interacted with the protein through hydrogen bonding.Residues such as Arg 79, Phe 138, Ser 139 and Asp 273 interacted mostly with all the drugs (Table 2).These residues made hydrophobic interactions with the ligands (except the -OH group of Ser 139 that interacted with 3-amino-N-carbamimidoyl-6-chloro-5- In Vc-NhaP2, Asp 273 located in TMS X, is involved in maintaining the ion selectivity of the antiporter [16,18] since mutating it to Ala 273 makes the antiporter only selective for K + whereas no activity is observed with Na + [16,18].Ala 135 in TMS V and Glu 155 in TMS VI are also involved in interactions with the ligands (Figure 4, Table 2).Interestingly, Ala 135 is present near the TD motif (Thr 132 and Asp 133) [16,18]; the TD motif is essential for the functioning of the Vc-NhaP2 antiporter because this motif is present in the extended region of TMS V and is involved in direct cation binding [16,18].We showed that Glu 155 in TMS VI is directly involved in forming the cation-binding pocket along with the other amino acids present in TMS VI (e.g., Asn 161, Glu 162 and Ser 156) [16,18].These interactions between the amino acid residues present in the protein and the ligands would likely interfere with the function of TMSs V, VI and TMS X, since these TMSs are located in close proximity in the threedimensional structure of the protein (Supplementary Figure 1).interact with all the three ligands (Figure 4).Among them Val 170, Phe 236, Ile 247 and Gly 244 make hydrophobic interactions.The OH group of Ser 245 forms H-bonds with 3,5-diamino-6chloro-N-cyanopyrazine-2-carboxamid, Amikol 1 and 2. Asn 66 and Asn 240 make both hydrophobic and H-bonding with the ligands.However, Asn 240 uses the carbonyl group of the backbone to create a H-bond with Amikol 2. These amino acid residues are conserved in the other NhaP-type antiporters [15] and are present in TMSs IV, VI, X, XI, XII and XIII.The biological functions of these amino acid residues are still not clear.

Root Mean Square Deviation (RMSD) of backbone C-α carbon
MD simulations were carried out with the four selected docked complex structures and compared to the apo-Vc-NhaP2 structure to measure the dynamics and stabilities of the drugbound complexes (Figures 5 and 6).The dynamics of the protein were probed by determination of the RMSD's of backbone C-α carbon atoms for the apoas well as the drug-bound-Vc-NhaP2 complexes (Figure 5).The RMSD values indicate that the protein stability of the drug-bound complexes deviates from that of the apo-Vc-NhaP2 on the 1 microsecond timescale (Figure 5).

Root Mean Square Fluctuation (RMSF) backbone C-α carbon
Whereas the overall dynamics of the protein and the drug complexes are only slightly different a few areas of the protein show greatly increased dynamics upon drug binding.This is indicated in Figure 6 where the mobility and flexibility of the backbone C-α carbons of each amino acid residue is indicated from the root mean square fluctuation (RMSF) analysis (Figure 6).In the apo-Vc-NhaP2 higher fluctuations are noticeable in TMS IV (amino acid residue from 81V to 111L), TMS V (118L to 142 G) and TMS VII (333K to 350A, TMS XIII (362N to 379N) (Figure 6).The flexibility in TMSs V and VII is crucial for the ion transport of Vc-NhaP2 by the alternating-access mechanism [16], whereas TMS VI, containing the ion binding residues, remains rigid.Upon binding of Amiko-1, it greatly elevated dynamics are observed in TMS VI (149V to 176G) compared to the apo-Vc-NhaP2 (Figure 6).The largest increase in the mobility of the amino acid residues (RMSF >0.3 nm) in TMS V and VI are visible in the Amikol 1-bound structure (Figure 6).These results clearly indicate an effect of the drug on the dynamics of the amino acid residues present in the functionally-important TMSs.Bad Bonds 0

Discussion
The rising incidence of multidrug-resistant V. cholerae outbreaks is an emerging issue in developing countries [1,3].A great deal of research has been done to investigate the role of resistance-inducing multidrug efflux antiporters whereas cation-proton antiporters have not been considered possible targets for drug targeting [55][56].Since cation-proton antiporters play a major role in the survival of the bacteria, blocking their activity with suitable drugs could be a prospective therapy [16].
As we reported, the membrane of V. cholerae contains a trio of cation-proton antiporters of a specific type, NhaP, that are responsible for the transport of both K + and Na + [12][13][14][15][16][17][18].They are encoded by three paralogous structural genes, Vc-nhaP1, 2 and 3. Our phenotypical analysis of deletion mutants suggested that these antiporters play an essential role in bacterial physiology [12][13][14][15][16][17][18].We hypothesized that they comprise a novel mechanism of the Acid Tolerance Response (ATR).Preliminary ATR tests conducted with the wild-type parental strain V. cholerae O395N1 and its triple deletion mutant Vc∆NhaP123, showed that the triple mutant died at a much higher rate than the isogenic wild type strain upon being challenged by HCl (pH 3.5), [13].Considering the possible physiological role of Vc-NhaP isoforms in the ATR of V.
cholerae, we hypothesized that the inhibition of Vc-NhaP paralogues might inhibit the infectious process caused by this pathogen when it crosses the gastric acid barrier.Since Vc-NhaP2 could be a potential target for future drug development targeting these ion exchangers, inhibition of this protein alone might impair the overall infectious process caused by V. cholerae as it passes though the gastric acid barrier to initiate the infection [13,16].
The NhaP-type antiporters are phylogenetically diverse and present in both eukaryotes and prokaryotes [12].They also play a crucial role in enhancing the survival of different blood-borne pathogenic bacteria in human blood.For example, it was reported that survival of Yersinia pestis in blood is related to the presence of NhaA and NhaB sodium-proton antiporters.Deletion of these Na + /H + antiporters decreased its survival in blood [57].Developing inhibitors against the Vc-NhaP2 antiporter might also be broadly applicable to the antiporters present in the genomes of pathogenic microorganisms [19].
For this study we selected Amiloride, a potent inhibitor of the human Na + /H + exchanger (NHE-1), as a lead compound to target Vc-NhaP2.After extensive exploration of ADMET and QSAR properties, we used in silico methods to determine the pharmacological consequences of AMLRD and its analogues (Supplementary file 1).Among them, six AMLRD analogs with modest pharmacokinetic features and the lowest toxicities were obtained for molecular docking simulations against Vc-NhaP2.The selected compounds demonstrate average binding affinity scores around -6.0 kcal/mol.Docking studies clearly indicate two potential binding sites; either at the cytoplasmic side or at the periplasmic side of the protein (Table 2).Moreover, amino acid residues located in TMSs IV, V, VI and the loop between VIII-XII are involved in drug binding.
Importantly, TMS V, VI and XII are crucial for the functioning of the Vc-NhaP2 antiporter, as amino acid residues present on these TMSs are directly involved in cation binding [16,18].
Toxicity is a major issue during drug development.Around 33% of drug candidates cannot pass all the clinical trials due to their toxicity and result in significant economic costs [58].AMLRD and its analogs contain chemical properties that can lead to hepatoxicity and mutagenicity (Supplementary file 1) [59].To overcome these problems, the results of pkCSM, admetSAR 2.0, OSIRIS property explorer and ProTox-II were used to evaluate drug toxicities.
Especially, data generated by ProTox-II were evaluated meticulously since this virtual lab was established to decrease animal model experiments by integrating robust prediction models on drug toxicity [32].These analyses indicated that some of the selected compounds have are potentially toxic thus rational drug design was conducted to lessen the toxicity of particular drug candidates (Table 1).However, some designed molecules such as AmiKol 2 and 3 still contains irritant characteristics though Amikol 3 has reduced hepatotoxic properties.To ensure proper safety, these drugs require in vitro and in vivo validations.These drugs with lower toxicity also demonstrated nearly the same binding scores as their parent compound showed against the Vc-NhaP2.Lastly, the final drug candidates have minimum toxicity and maintain Lipinski's rule of five (Table 3) [60].
The four optimal NhaP2-drug complexes were elected for MD simulation where these complexes were embedded in a lipid bilayer membrane mimicking the native membrane of V.
cholerae [49].MD simulations for apo-Vc-NhaP2 revealed that the positions of the amino acid residues forming the putative cation binding pocket present in TMSs VI did not show any significant fluctuations compared to the residues present in other TMSs V and XII (Figure 6).
We previously suggested that the conformational changes occurring in TMSs V and XII in Vc-NhaP2 alternatively opens the rigid ion binding pocket in TMS VI either to the cytoplasm or to the periplasmic side [16].In the present study we have hypothesized that binding of our drug candidates would possibly affect the flexibility or the rigidity of crucial TMSs of Vc-NhaP2 destabilizing the protein and hence altering its function.Our findings clearly suggest that drugbound structures have higher flexibility in the amino acid residues present in the TMS VI involved in ion binding (Figure 5).This observation indicates that the binding of drugs could plausibly affect the flexibility of the Vc-NhaP2 which is essential for the functioning of the antiporter.In addition, upon drug binding the global structure of Vc-NhaP2 deviated from the unbound structure (Figure 5).

Molecular docking between Vc-NhaP2 and drug candidates
To discover the interactions between Vc-NhaP2 and the six selected drug candidates, molecular docking was executed by AutoDock Vina (ADV) [41].Here, Vc-NhaP2 was selected as antiporter and the nominated AMLRD analogues were designated as ligands.Since, there is no prior knowledge about the drug binding site in V. cholerae NhaP2 a blind docking approach was employed [42].The number of torsion angles was kept at zero in the ligands and the polar hydrogen bonds were added in the protein.ADV is a renowned software for molecular docking that gave probable binding sites of the drugs with a binding score The binding site with the best binding score was taken for further analysis.In addition, the Achilles Blind Docking Server (https://bio-hpc.ucam.edu/achilles/)was implemented for additional validation.This web-based server has been used in different academic and industrial purposes for blind molecular docking.
The best binding scores for each protein-drug complex derived from the Achilles Blind Docking server were also collected.ADV deduced interactions between the drugs and Vc-NhaP2 were visualized by PyMol [35], UCSF Chimera [43] and LigPlot + [44].

Rational drug design
Based on the best potential binding score of the protein-ligand complexes (Table 2) and toxicity profiles (Table 1), two drugs (3,5-diamino-6-chloro-N-cyanopyrazine-2-carboxamide, PubChem CID: 123478999 and 3-amino-N-carbamimidoyl-6-chloro-5(dimethylamino)pyrazine-2-carboxamide, PubChem CID: 137630036) were selected for modification in order to reduce their toxicity and mutagenicity.To do so, Structure Activity Relationship (SAR) of AMLRD were explored by reviewing the literature [45][46].Then, SAR helped to identify the side-chains or functional groups to be manipulated.Secondly, drug-antiporter complexes were examined thoroughly after molecular docking.This facilitated the detection of the atoms that were critical during ligand-protein interactions.Following this, the compounds were modified via Molinspiration (https://www.molinspiration.com/)(Figure 1).

Energy minimization of the modified drugs
Computationally-designed drugs require energy minimization to improve physical realism, side-chain accuracy and stereochemistry [47][48].To perform energy minimization, the PDB files of the modified drugs were input into the YASARA (Yet Another Scientific Artificial Reality Application) Energy Minimization Server.YASARA performed knowledge-based forcefield energy minimization [48].The energy-minimized structures were then docked to the refined protein structure.

Figure 1 .
Figure 1.Nominated Amiloride analogues for blocking the Vc-NhaP2 antiporter.(A) A, B and

Figure 3 .
Figure 3. Z-score plot and plot of residue scores of the Vc-NhaP2 structure generated by Pro-SA

Figure 5 .
Figure 5. Time-dependent root mean square deviation (RMSD) of backbone C-α of the apo-Vc-

Figure 6 .
Figure 6.The root-mean squared fluctuations (RMSF) for the C-α atoms of the backbone of the

Table 1 .
Absorption, distribution, metabolism, excretion and toxicity analysis of the selected and

Table 2 .
Interaction of drugs and Nhap2 with participated amino acid residues

Table 3 :
Physicochemical properties of the finally selected drug

Table 4 .
MolProbity results of refined Vc-NhaP2 protein structure as predicted by SWISS-